Diurnal Air–Surface Temperature Dynamics in Hong Kong: An Analysis Across Land Use and Climate Zones

Understanding the relationship between air temperature (T_a) and surface temperature (T_s) is crucial for urban climate studies, as it aids sustainable urban development and climate adaptation. Given the critical role of urban morphology in climate dynamics, this study examines how urban configurations impact temperature variations as well as provides evidence for future development and mitigating heat-related risks. We examined the relationship between T_a and T_s through the lens of urban configuration, considering land use (LU) and local climate zones (LCZs) for the densely populated city of Hong Kong. Additionally, multiple linear regression (MLR) and random forest (RF) models were developed to predict spatially continuous T_a using T_s, LU, LCZ and time of day. Our analysis revealed a dynamic relationship between thermal dynamics and both time of day as well as urban configurations, with a stronger relationship between T_a and T_s in the daytime than at night. Furthermore, the difference between T_a and T_s showed a stronger correlation with LCZ than with LU classes. Model evaluation demonstrated the superior performance of RF over MLR in predicting spatially continuous T_a (R² = 0.80 vs. 0.75). The results from SHapley Additive exPlanations (SHAP) model analysis of RF prediction further emphasised the substantial contribution of residential LU and compact high-rise LCZ in continuous T_a prediction in the study area. These findings offer valuable insights for urban planning in densely populated cities, aiding in the formulation of development plans that promote stable urban thermal comfort.